A safety identification method for high-altitude work protection in tunnels

Abstract

The present invention relates to the field of tunnel construction safety protection technology, and discloses a method for identifying safety protection for high-altitude work in tunnels. The method comprises acquiring tunnel face construction image data; constructing and pre-training a high-altitude work target recognition model in the tunnel; wherein the high-altitude work targets in the tunnel include a trolley, a worker, and a guardrail; utilizing the trained high-altitude work target recognition model in the tunnel to obtain the identification type and location information of the high-altitude work targets in the tunnel based on the tunnel face construction image data; determining the target spatial relationship based on the identification type and location information of the high-altitude work targets in the tunnel; determining the high-altitude work scene in the tunnel based on the spatial relationship between the trolley and the worker; and determining the high-altitude work protection safety status based on the spatial relationship between the worker and the guardrail, based on the high-altitude work scene in the tunnel. The present invention achieves accurate identification of the high-altitude work process in the tunnel face and the high-altitude work protection safety status of the workers.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel construction safety protection, and in particular to a method for identifying safety of high-altitude operations protection in a tunnel. Background Art

[0002] Tunnel (road) projects, as a crucial component of infrastructure, play a vital role in transportation, hydropower development, and water resource utilization. During construction, tunnels often face complex and changing geological conditions, confined spaces, and harsh working environments. Construction workers are more vulnerable to accidents such as falls from heights, impacts, tunnel collapses, and mechanical collisions, resulting in significant loss of life and property.

[0003] Tunnel construction involves numerous coordinated operations, including manual drilling, charging, erecting frames, and secondary lining, all of which require operation at height. Unsafe worker behavior and a lack of appropriate safety equipment are the primary causes of falls from height. The status of high-altitude protective equipment on construction sites is typically determined through manual inspections and analysis of video surveillance footage, which is often time-consuming and labor-intensive, and makes it difficult to ensure timely inspections. Furthermore, tunnel projects are often long, linear projects, and as excavation deepens, construction conditions become increasingly complex. Manual inspections often have low visibility and blind spots, making it impossible to conduct a comprehensive inspection of construction safety.

[0004] With the advancement of sensor and computer vision technologies, automated, real-time monitoring of the safety status of workers working at heights has become a reality. However, sensor-based methods require workers to wear additional sensing equipment, which is expensive and disrupts their normal operations. Computer vision methods are low-cost and minimally impactful to workers, but they face challenges such as poor lighting conditions in tunnels, complex image backgrounds, and variations in target size. Therefore, it is necessary to develop a safety identification method for high-altitude work safety, suitable for the complex environments of tunnels, to enable rapid identification and feedback of safety issues. Summary of the Invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a safety identification method for high-altitude work protection in a cave.

[0006] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0007] A method for identifying safety during high-altitude work in a cave comprises the following steps:

[0008] Acquire tunnel face construction image data;

[0009] Constructing a model for identifying targets of high-altitude operations in tunnels and performing pre-training; the targets of high-altitude operations in tunnels include trolleys, workers, and guardrails;

[0010] The trained in-tunnel aerial work target recognition model is used to obtain the in-tunnel aerial work target recognition type and location information based on the tunnel face construction image data;

[0011] Determine the target spatial relationship based on the identification type and location information of the high-altitude operation target in the cave;

[0012] Determine the high-altitude operation scenario in the tunnel based on the spatial relationship between the trolley and the workers;

[0013] Based on the high-altitude work scene in the cave, the high-altitude work protection safety status is determined according to the spatial relationship between the workers and the guardrails.

[0014] Preferably, the pre-training of the indoor high-altitude operation target recognition model includes:

[0015] Use the CycleGAN network to enhance tunnel face construction images;

[0016] Annotate the enhanced tunnel face construction image with examples.

[0017] Preferably, using the CycleGAN network to enhance the tunnel face construction image includes:

[0018] The first discriminator judges the input low-light tunnel face construction image to determine whether it meets the characteristics of images in the low-light tunnel dataset and outputs a probability value indicating that the input image is a true low-light image;

[0019] The input low-illumination tunnel face construction image is input into the first generator to generate a normal-illumination tunnel face construction image. The generated image is then judged by the second discriminator to determine whether it meets the characteristics of images in the normal-illumination tunnel dataset and output a probability value indicating that the input image is a true normal-illumination image.

[0020] The normal-illumination tunnel face construction image generated by the first generator is input into the second generator to regenerate the low-illumination tunnel face construction image, so that the generated low-illumination tunnel face construction image is consistent with the input low-illumination tunnel face construction image, thereby forming an image conversion cycle;

[0021] Calculate the adversarial loss, cycle consistency loss, and identity mapping loss in the image conversion process, use the optimizer to calculate the updated loss function gradient value and the parameter values ​​of the generator and discriminator, and iterate continuously through the gradient descent method until the overall loss function is minimized.

[0022] Preferably, the target recognition model for high-altitude operations in a cave includes:

[0023] Backbone network, neck network and prediction network;

[0024] The backbone network includes a first CBS module, a first C3 module, a second CBS module, a second C3 module and an SPPF module arranged in sequence, as well as a first attention module arranged at the output end of the second CBS module, a second attention module arranged at the output end of the second C3 module, and a third attention module arranged at the output end of the SPPF module.

[0025] Preferably, the first attention module, the second attention module and the third attention module each include:

[0026] Height global average pooling unit, width global average pooling unit, stacking unit, first convolution unit, normalization unit, first activation unit, split unit, transposition unit, second convolution unit, third convolution unit, second activation unit, third activation unit and multiplication unit;

[0027] The height global average pooling unit and the width global average pooling unit respectively perform global average pooling operations in the height direction and the width direction on the input feature map to obtain a first global average pooling feature map and a second global average pooling feature map;

[0028] The stacking unit performs a stacking operation on the first global average pooling feature map and the second global average pooling feature map, and then sequentially passes through a first convolution unit, a normalization unit, and a first activation unit to obtain a first spatial feature map;

[0029] After the segmentation unit separates the width direction feature map and the height direction feature map from the first spatial feature map, the transposition unit performs transposition operations on the width direction feature map and the height direction feature map respectively, and then passes the two feature maps through the second convolution unit and the third convolution unit, as well as the second activation unit and the third activation unit respectively, to obtain width direction attention and height direction attention;

[0030] The multiplication unit multiplies the input feature map with the width direction attention and the height direction attention to obtain an attention feature map.

[0031] Preferably, determining the target spatial relationship based on the identification type and position information of the high-altitude operation target in the cave includes:

[0032] Determine the spatial inclusion, intersection, or separation relationship between the worker and trolley target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the trolley detection box;

[0033] The spatial inclusion, intersection, or separation relationship between the worker and guardrail target category bounding boxes is determined based on the coordinate information of the worker detection box and the coordinate information of the guardrail detection box.

[0034] Preferably, determining the spatial inclusion, intersection, or separation relationship between the worker and trolley target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the trolley detection box includes:

[0035] When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then there is a spatial inclusion relationship between the worker and the trolley target category bounding box;

[0036] When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then there is a spatial intersection relationship between the worker and the trolley target category bounding box;

[0037] When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then the worker and the trolley target category boundary box are spatially separated.

[0038] Preferably, determining the spatial inclusion, intersection, or separation relationship between the worker and guardrail target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the guardrail detection box includes:

[0039] When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then there is a spatial inclusion relationship between the worker and the guardrail target category bounding box;

[0040] When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then there is a spatial intersection relationship between the worker and the guardrail target category bounding box;

[0041] When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then the worker and the guardrail target category bounding box are spatially separated.

[0042] Preferably, determining the high-altitude working scene in the tunnel based on the spatial relationship between the trolley and the worker includes:

[0043] If the worker and the trolley target category bounding boxes are spatially contained or spatially intersected, the scene is judged to be an indoor high-altitude operation process;

[0044] If only the trolley target is detected, only the worker target is detected, or both are detected and the trolley and all the worker target category bounding boxes are spatially separated, it is judged as a non-high-altitude operation process.

[0045] Preferably, based on the high-altitude working scene in the cave, determining the high-altitude working protection safety status according to the spatial relationship between the worker and the guardrail includes:

[0046] According to the shooting range of the surveillance camera and the size of the trolley, the set area above the trolley detection frame is delineated as the high-altitude working area, and the spatial relationship between the worker and the guardrail is judged within this area. If the high-altitude working worker detection frame is spatially separated from a certain guardrail detection frame, it is judged that the worker is in a dangerous working state with the guardrail missing; if the high-altitude working worker detection frame is spatially intersected or spatially contained with all guardrail detection frames, it is judged that the worker is in a safe working state.

[0047] The present invention has the following beneficial effects:

[0048] 1. The present invention performs image enhancement on images inside the tunnel. The processed images have significant improvements in four indicators: brightness, average gradient, energy gradient, and information entropy, which enhances the visual effect of the images and improves the detection accuracy of subsequent image recognition methods.

[0049] 2. The present invention realizes the automatic processing and recognition of tunnel construction image data, and based on the spatial relationship of identified targets such as workers, trolleys, and guardrails, it realizes the accurate identification of high-altitude work processes on the face and the safety status of workers' high-altitude work protection, providing technical support for all-weather, all-round and efficient safety management of high-altitude edge areas in the tunnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of a safety identification method for high-altitude work protection in a cave;

[0051] Figure 2 Schematic diagram of the CycleGAN network structure for preprocessing images inside the cave;

[0052] Figure 3 This is a schematic diagram of the target recognition model structure for high-altitude operations in tunnels;

[0053] Figure 4 This is a schematic diagram showing an example of the spatial relationship between the trolley and the worker;

[0054] Figure 5 Schematic diagram of the confusion matrix for high-altitude scene recognition;

[0055] Figure 6 Schematic diagram of the confusion matrix for protection status identification;

[0056] Figure 7 This is a schematic diagram showing an example of the spatial relationship between workers and guardrails;

[0057] Figure 8 Schematic diagram of the safety identification results for workers working at heights. DETAILED DESCRIPTION

[0058] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0059] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying safety during high-altitude operations in a cave, comprising the following steps S1 to S6:

[0060] S1. Acquire tunnel face construction image data;

[0061] In an optional embodiment of the present invention, step S1 uses the video data obtained from the tunnel face video monitoring to capture the construction images inside the tunnel. In order to adapt to the characteristics of tunnel engineering such as changeable lighting conditions and complex construction environment inside the tunnel, a camera with high definition, wide angle, dustproof and explosion-proof functions is selected.

[0062] S2. Build a model for identifying targets of high-altitude operations in tunnels and perform pre-training. The targets of high-altitude operations in tunnels include trolleys, workers, and guardrails.

[0063] In an optional embodiment of the present invention, when step S2 pre-trains the target recognition model for high-altitude operations in the tunnel, real tunnel construction scenes from different distances, angles, locations, and processes are screened out from the images obtained from the camera, including construction workers, trolleys, guardrails in high-altitude operation scenes such as drilling, charging and blasting, and construction workers in non-high-altitude operation scenes. A total of 690 images are collected to construct a data set for feature analysis of construction images in the tunnel.

[0064] This embodiment pre-trains the model for identifying targets for high-altitude operations in a cave, including:

[0065] Use the CycleGAN network to enhance tunnel face construction images;

[0066] Annotate the enhanced tunnel face construction image with examples.

[0067] Due to the poor lighting conditions in construction tunnels, video surveillance images are affected by smoke and dust inside the tunnel, and have characteristics such as low illumination and low definition. In this embodiment, CycleGAN is used to enhance the tunnel surveillance images to improve the subsequent detection accuracy. CycleGAN uses adversarial loss function, cycle consistency loss function and identity mapping loss function to perform constrained optimization on the target image data, thereby converting the original low-illumination image domain features into normal illumination image domain features to the greatest extent possible. The model network structure is as follows: Figure 2 As shown, the total loss function and the overall optimization objective function formula are as follows:

[0068]

[0069]

[0070] in, For the generator , For the generator , For the discriminator , For the discriminator , is the total loss function, To counter the loss function, is the cycle consistency loss function, is the identity mapping loss function, is the weight of cycle consistency loss, is the weight of the identity mapping loss, is the overall optimization objective function.

[0071] CycleGAN includes forward conversion and reverse conversion processes, and the two conversion processes are carried out simultaneously. Taking the forward conversion process as an example, the specific steps are as follows:

[0072] (1) Through the first discriminator The input low-light tunnel image is judged to determine whether it meets the characteristics of the images in the low-light tunnel dataset, and a probability value is output to represent the probability that the input image is a true low-light image.

[0073] (2) Input the low-light tunnel image into the first generator Generate a tunnel image with normal illumination, and then pass it through the second discriminator The generated image is judged to determine whether it meets the characteristics of the images in the normal illumination tunnel dataset, and a probability value is output to represent the probability that the input image is a true normal illumination image.

[0074] (3) The first generator The normal illumination tunnel image generated in the second generator is input In the process, the low illumination tunnel image is regenerated so that the generated low illumination tunnel image is as consistent as possible with the input low illumination tunnel image, forming a cycle.

[0075] (4) Calculate the adversarial loss, cycle consistency loss, and identity mapping loss in the image conversion process, use the optimizer to calculate the updated loss function gradient value and the parameter values ​​of the generator and discriminator, and iterate continuously through the gradient descent method until the overall loss function is minimized.

[0076] After preprocessing, the LabelImg annotation tool is used to annotate the worker, trolley, and guardrail instances in each image in the dataset, thereby forming an image sample for identifying the safety status of high-altitude work protection in the tunnel.

[0077] Using pre-processed tunnel face surveillance video, we generated image samples demonstrating the safety status of high-altitude work protection within the tunnel. These images were divided into a training dataset and a test dataset (each image was annotated with labels such as the type of vehicle, worker, and guardrail). The image dataset served as the model input, and the categories and locations of the three detected objects served as the model output.

[0078] In an optional embodiment of the present invention, the in-hole aerial work target recognition model constructed in step S2 includes:

[0079] Backbone network, neck network and prediction network;

[0080] The backbone network includes a first CBS module, a first C3 module, a second CBS module, a second C3 module and an SPPF module arranged in sequence, as well as a first attention module arranged at the output end of the second CBS module, a second attention module arranged at the output end of the second C3 module, and a third attention module arranged at the output end of the SPPF module.

[0081] The first attention module, the second attention module, and the third attention module all include:

[0082] Height global average pooling unit, width global average pooling unit, stacking unit, first convolution unit, normalization unit, first activation unit, split unit, transposition unit, second convolution unit, third convolution unit, second activation unit, third activation unit and multiplication unit;

[0083] The height global average pooling unit and the width global average pooling unit respectively perform global average pooling operations in the height direction and the width direction on the input feature map to obtain a first global average pooling feature map and a second global average pooling feature map;

[0084] The stacking unit performs a stacking operation on the first global average pooling feature map and the second global average pooling feature map, and then sequentially passes through a first convolution unit, a normalization unit, and a first activation unit to obtain a first spatial feature map;

[0085] After the segmentation unit separates the width direction feature map and the height direction feature map from the first spatial feature map, the transposition unit performs transposition operations on the width direction feature map and the height direction feature map respectively, and then passes the two feature maps through the second convolution unit and the third convolution unit, as well as the second activation unit and the third activation unit respectively, to obtain width direction attention and height direction attention;

[0086] The multiplication unit multiplies the input feature map with the width direction attention and the height direction attention to obtain an attention feature map.

[0087] The network structure of the in-hole aerial work target recognition model constructed in this embodiment is to add an attention module after the CBS module, C3 module and SPPF module of the original YOLOv5 model backbone network. The network structure is as follows: Figure 3 As shown. First, according to the detection task requirements and computing resource constraints, YOLOv5s is selected as the detection model, and the CA attention mechanism is introduced into the detection model to improve the detection accuracy. Then, the YOLOv5-CA model is trained using the divided training set, and the hyperparameters are set as follows: the input image size is 640×640, the batch training data size is set to 32, the initial learning rate is set to 0.001, the weight decay coefficient is set to 0.0005, the training momentum is set to 0.937, the training rounds are set to 400, and the optimizer is Adam. After multiple iterative training, the model network structure, loss function and optimizer parameters are continuously adjusted to obtain the YOLOv5-CA target detection model for construction workers, trolleys and guardrails in the tunnel construction scene with optimal parameters. Finally, the model detection performance and generalization ability are evaluated based on the divided test set. Precision, recall and mean average precision (mAP) are selected as evaluation indicators to evaluate the accuracy of the model in detecting workers, trolleys and guardrails. The calculation formula is as follows:

[0088]

[0089]

[0090]

[0091] Where, 、 They are precision and recall respectively; true positives (TP) are the number of correctly detected targets with an intersection over union (IoU) greater than a set threshold (usually 0.5); false positives (FP) are the number of actual non-targets detected; false negatives (FN) are the number of undetected targets; The number of types of PPE tested.

[0092] At this point, the model achieved precision, recall, and mean average precision (mAP) of 97.8%, 93.3%, and 96.0%, respectively, representing a 1.69% improvement compared to the 94.4% mAP of the pre-improvement model. The model was able to accurately identify trolleys, workers, and guardrails in tunnel construction images, demonstrating its applicability for identifying targets involved in aerial work within tunnels.

[0093] S3, using the trained in-tunnel high-altitude operation target recognition model to obtain the in-tunnel high-altitude operation target recognition type and location information based on the tunnel face construction image data;

[0094] In an optional embodiment of the present invention, step S3 uses the in-hole aerial work target recognition model established in step S2 to identify the coordinates of the worker detection frame A. , coordinates of the trolley detection frame B And the coordinates of the guardrail detection frame C .

[0095] S4. Determine the target spatial relationship based on the identification type and position information of the high-altitude operation target in the cave;

[0096] In an optional embodiment of the present invention, step S4 of determining the target spatial relationship based on the identification type and position information of the aerial work target in the cave includes:

[0097] Determine the spatial inclusion, intersection, or separation relationship between the worker and trolley target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the trolley detection box;

[0098] The spatial inclusion, intersection, or separation relationship between the worker and guardrail target category bounding boxes is determined based on the coordinate information of the worker detection box and the coordinate information of the guardrail detection box.

[0099] This embodiment is based on the coordinates of the worker detection frame A , coordinates of the trolley detection frame B And the coordinates of the guardrail detection frame C ,determine the spatial inclusion, intersection or separation relationship between the bounding boxes of the worker and the trolley, and the worker and the guardrail.

[0100] The spatial inclusion, intersection, or separation relationship between the worker and trolley target category bounding boxes is determined based on the coordinate information of the worker detection box and the coordinate information of the trolley detection box, including:

[0101] When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet ,Right now:

[0102]

[0103] in Indicates the coordinates of the upper left corner of the worker detection box, Indicates the coordinates of the lower right corner of the worker detection box, Indicates the coordinates of the upper left corner of the trolley detection frame, Indicates the coordinates of the lower right corner of the vehicle detection frame.

[0104] There is a spatial inclusion relationship between the worker and the trolley target category bounding box.

[0105] When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet ,Right now:

[0106]

[0107] There is a spatial intersection relationship between the worker and the trolley target category bounding boxes.

[0108] When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then the worker and the trolley target category boundary box are spatially separated.

[0109] Determining the spatial inclusion, intersection, or separation relationship between the worker and guardrail target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the guardrail detection box includes:

[0110] When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet ,Right now:

[0111]

[0112] Then there is a spatial inclusion relationship between the worker and the guardrail target category bounding box;

[0113] When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet ,Right now:

[0114]

[0115] Then there is a spatial intersection relationship between the worker and the guardrail target category bounding box;

[0116] When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then the worker and the guardrail target category bounding box are spatially separated.

[0117] S5. Determine the high-altitude operation scene in the tunnel based on the spatial relationship between the trolley and the workers;

[0118] In an optional embodiment of the present invention, step S5 determines whether it belongs to an aerial work scene based on the spatial relationship between the trolley detection frame and the worker detection frame, specifically:

[0119] If the worker and the trolley target category bounding boxes are spatially contained or spatially intersected, the scene is judged to be an indoor high-altitude operation process;

[0120] If only the trolley target is detected, only the worker target is detected, or both are detected and the trolley and all the worker target category bounding boxes are spatially separated, it is judged as a non-high-altitude operation process.

[0121] by Figure 4 Taking the above as an example, we conduct a specific analysis of the spatial relationship between the trolley detection frame and the worker detection frame to determine whether it belongs to a high-altitude operation scene. Figure 4 The middle trolley and workers 2, 3, 4, 5, 6, 7, 8, and 9 are all spatially contained. To achieve integrated construction of the tunnel face, the trolley's dimensions are often similar to the tunnel cross-section. Workers working at high altitude near the edge are spatially contained within the trolley, and this is considered an aerial work process. Figure 4 The trolley and worker 1 are in a spatially intersecting relationship, meaning they may intersect when working on top of the trolley or on the ground, and can also be considered as working in an aerial scene. When only the trolley or the worker are detected, or the worker and trolley are very far apart (due to the limited viewing angle of the image, the situation where the person and trolley coexist but are very far apart generally does not occur), they are in a spatially separated relationship and are considered non-aerial work processes.

[0122] 87 tunnel images were randomly selected as the data set for model verification, of which 36 images belong to high-altitude work processes and 51 images belong to non-high-altitude work processes. The data set was input into the model, and the confusion matrix obtained was as follows Figure 5 As shown, it can be seen that there are 33 high-altitude work processes and 54 non-high-altitude work processes, among which TP is 32, FP is 1, and FN is 4. That is, the Precision and Recall of the model are 96.97% and 88.89% respectively, which has a high recognition accuracy.

[0123] S6. Based on the high-altitude work scene in the tunnel, determine the high-altitude work protection safety status according to the spatial relationship between the worker and the guardrail.

[0124] In an optional embodiment of the present invention, step S6 continues to identify the safety protection status of the high-altitude workers based on the high-altitude work scene identified in step S5. Based on the shooting range of the surveillance camera and the size of the trolley, a certain area above the trolley detection frame is demarcated as the high-altitude work area (this value depends on the size of the tunnel), and the spatial relationship between the workers and the guardrails within this area is analyzed to identify the safety protection status of the high-altitude workers. If the high-altitude worker detection frame and a certain guardrail detection frame are spatially separated, the worker is judged to be in a dangerous working state with a missing guardrail; if the high-altitude worker detection frame and all guardrail detection frames are spatially intersected or spatially contained, the worker is judged to be in a safe working state.

[0125] by Figure 7 Taking the example of the above, the spatial relationship between the worker detection frame and the guardrail detection frame is specifically analyzed to determine whether it belongs to a safe working state. Figure 7 In the figure, workers 1, 2, 3, and 4 all have a spatial inclusion relationship with the guardrail. When workers are working at height, the presence of a guardrail near their edge can effectively prevent them from falling. This spatial relationship indicates that workers are protected by the guardrail when squatting or bending over, indicating a safe working state. Figure 7 The middle guardrail and worker 6 are in a spatial intersection relationship, indicating that the worker is protected by the guardrail when working upright, which is a safe working state; when there is no guardrail in front of the worker in the high-altitude working area, it is a spatial separation relationship, and the worker is in a dangerous working state.

[0126] After the image is judged as a high-altitude work process, the worker's safety status is further judged. Among the 32 TP samples, there are a total of 267 worker instances, 144 of which are in a safe state and 123 are in a dangerous state. Figure 6 From the output confusion matrix, we can see that the model output results are 146 people in a safe state and 121 people in a dangerous state, among which TP is 140, FP is 6, and FN is 4. That is, the Precision and Recall of the model are 95.89% and 97.22% respectively, and the accuracy meets the engineering requirements. Figure 8 This is part of the model's output for safety identification of workers working at height. The green detection box indicates that the worker is in a safe state, and the red detection box indicates that the worker is in a dangerous edge state.

[0127] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0128] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0129] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0130] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

[0131] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A safety identification method for high-altitude work protection in a cave, characterized in that: The following steps are involved: Acquire tunnel face construction image data; Build a target recognition model for high-altitude operations in caves and perform pre-training; The objects of the high-altitude operation in the cave include trolleys, workers and guardrails; Pre-training of the indoor high-altitude operation target recognition model includes: The CycleGAN network is used to enhance tunnel face construction images, including: The first discriminator judges the input low-light tunnel face construction image to determine whether it meets the characteristics of images in the low-light tunnel dataset and outputs a probability value indicating that the input image is a true low-light image; The input low-illumination tunnel face construction image is input into the first generator to generate a normal-illumination tunnel face construction image. The generated image is then judged by the second discriminator to determine whether it meets the characteristics of images in the normal-illumination tunnel dataset and output a probability value indicating that the input image is a true normal-illumination image. The normal-illumination tunnel face construction image generated by the first generator is input into the second generator to regenerate the low-illumination tunnel face construction image, so that the generated low-illumination tunnel face construction image is consistent with the input low-illumination tunnel face construction image, thereby forming an image conversion cycle; Calculate the adversarial loss, cycle consistency loss, and identity mapping loss during the image conversion process, use the optimizer to calculate and update the gradient value of the loss function and the parameter values ​​of the generator and discriminator, and iterate continuously through the gradient descent method until the overall loss function is minimized; Annotate the enhanced tunnel face construction image with examples; The trained in-tunnel aerial work target recognition model is used to obtain the in-tunnel aerial work target recognition type and location information based on the tunnel face construction image data; Determine the target spatial relationship based on the identification type and location information of the high-altitude operation target in the cave; Determine the high-altitude operation scenario in the tunnel based on the spatial relationship between the trolley and the workers; Based on the high-altitude work scene in the cave, the high-altitude work protection safety status is determined according to the spatial relationship between the worker and the guardrail, including: According to the shooting range of the surveillance camera and the size of the trolley, the set area above the trolley detection frame is delineated as the high-altitude working area, and the spatial relationship between the worker and the guardrail is judged within this area. If the high-altitude working worker detection frame is spatially separated from a certain guardrail detection frame, it is judged that the worker is in a dangerous working state with the guardrail missing; if the high-altitude working worker detection frame is spatially intersected or spatially contained with all guardrail detection frames, it is judged that the worker is in a safe working state.

2. A method for identifying safety during high-altitude work in a cave according to claim 1, characterized in that: The target recognition model for high-altitude operations in tunnels includes: Backbone network, neck network and prediction network; The backbone network includes a first CBS module, a first C3 module, a second CBS module, a second C3 module and an SPPF module arranged in sequence, as well as a first attention module arranged at the output end of the second CBS module, a second attention module arranged at the output end of the second C3 module, and a third attention module arranged at the output end of the SPPF module.

3. A method for identifying safety during high-altitude work in a cave according to claim 2, characterized in that: The first attention module, the second attention module and the third attention module each include: Height global average pooling unit, width global average pooling unit, stacking unit, first convolution unit, normalization unit, first activation unit, split unit, transposition unit, second convolution unit, third convolution unit, second activation unit, third activation unit and multiplication unit; The height global average pooling unit and the width global average pooling unit respectively perform global average pooling operations in the height direction and the width direction on the input feature map to obtain a first global average pooling feature map and a second global average pooling feature map; The stacking unit performs a stacking operation on the first global average pooling feature map and the second global average pooling feature map, and then sequentially passes through a first convolution unit, a normalization unit, and a first activation unit to obtain a first spatial feature map; After the segmentation unit separates the width direction feature map and the height direction feature map from the first spatial feature map, the transposition unit performs transposition operations on the width direction feature map and the height direction feature map respectively, and then passes the two feature maps through the second convolution unit and the third convolution unit, as well as the second activation unit and the third activation unit respectively, to obtain width direction attention and height direction attention; The multiplication unit multiplies the input feature map with the width direction attention and the height direction attention to obtain an attention feature map.

4. A method for identifying safety during high-altitude work in a cave according to claim 1, characterized in that: Determining the target spatial relationship based on the target identification type and location information of the high-altitude operation in the cave includes: Determine the spatial inclusion, intersection, or separation relationship between the worker and trolley target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the trolley detection box; The spatial inclusion, intersection, or separation relationship between the worker and guardrail target category bounding boxes is determined based on the coordinate information of the worker detection box and the coordinate information of the guardrail detection box.

5. A method for identifying safety during high-altitude work in a cave according to claim 4, characterized in that: The spatial inclusion, intersection, or separation relationship between the worker and trolley target category bounding boxes is determined based on the coordinate information of the worker detection box and the coordinate information of the trolley detection box, including: When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then there is a spatial inclusion relationship between the worker and the trolley target category bounding box; When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then there is a spatial intersection relationship between the worker and the trolley target category bounding box; When the coordinate information of the worker detection frame A and the coordinate information of the trolley detection frame B meet , then the worker and the trolley target category boundary box are spatially separated.

6. A safety identification method for high-altitude work protection in a cave according to claim 5, characterized in that: Determining the spatial inclusion, intersection, or separation relationship between the worker and guardrail target category bounding boxes based on the coordinate information of the worker detection box and the coordinate information of the guardrail detection box includes: When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then there is a spatial inclusion relationship between the worker and the guardrail target category bounding box; When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then there is a spatial intersection relationship between the worker and the guardrail target category bounding box; When the coordinate information of the worker detection frame A and the coordinate information of the guardrail detection frame C meet , then the worker and the guardrail target category bounding box are spatially separated.

7. A method for identifying safety during high-altitude work in a cave according to claim 1, characterized in that: Based on the spatial relationship between the trolley and the workers, the high-altitude working scenarios in the tunnel include: If the worker and the trolley target category bounding boxes are spatially contained or spatially intersected, the scene is judged to be an indoor high-altitude operation process; If only the trolley target is detected, only the worker target is detected, or both are detected and the trolley and all the worker target category bounding boxes are spatially separated, it is judged as a non-high-altitude operation process.

Images